14 



on these, the cows would be ci edited with too much milk, especially 

 where the flow of milk is decreasing-, as it usually is, except for the first 

 month or two after freshening". On the other hand, if weighings be made 

 op the iseventh and fourteenth days and the weights of milk for the pre- 

 ceding two weeks be calculated on the basis of these weights, the cow 

 would likely be credited with too little milk. By taking the weights on the 

 first, eighth, and fifteenth, and averaging these, we get probably as 

 nearly as can be to the actual amount of milk given by a cow in a two- 

 week period, weighing every seven days. 



The following table shows the actual error in a two-week period with 

 cows milked by hand regularly, and by cows usually milked with the 

 machine, but which were milked by hand and weighed daily for two weeks 

 in September, owing to a break of the motor which operates the vacuum 

 pump of the milking machine. 



Table Showing Error Introduced by Weighing Every Seven Days, as 

 Compared with Weighing Daily in September, igo6. 



By following the plan indicated of averaging the weights of three 

 daily milkings and multiplying by 14, to get the pounds of milk given in 

 two weeks, we find that the total error introduced in the case of four 

 cows regularly milked by hand, was 7 pounds milk. 



In the case of the second group of seven cows the total error in the 

 calculated yield of milk for the seven cows was but 9 pounds. The error 



